JANUARY 1999 PORCH ET AL. 69

Observations of Tracks from Ship-Based Platforms

W. P ORCH,* R. BORYS,ϩ P. D URKEE,# R. GASPAROVIC,@ W. H OOPER,& E. HINDMAN,** AND K. NIELSEN# * Los Alamos National Laboratory, Los Alamos, New Mexico ϩ Atmospheric Sciences Center, Desert Research Institute, Reno, Nevada # Department of , Naval Postgraduate School, Monterey, California @ The Johns Hopkins University, APL, Laurel, Maryland & Naval Research Laboratory, Washington, D.C. ** Earth and Atmospheric Sciences Department, City College of New York, New York, New York (Manuscript received 17 October 1997, in ®nal form 17 February 1998)

ABSTRACT Ship-based measurements in June 1994 provided information about ship-track and associated atmo- spheric environment observed from below levels that provide a perspective different from satellite and aircraft measurements. Surface measurements of latent and sensible heat ¯uxes, sea surface temperatures, and meteorological pro®les with free and tethered balloons provided necessary input conditions for models of ship- track formation and maintenance. Remote sensing measurements showed a coupling of ship plume dynamics and entrainment into overlaying clouds. Morphological and dynamic effects were observed on clouds unique to the ship tracks. These morphological changes included lower cloud bases early in the ship-track formation, evidence of raised cloud bases in more mature tracks, sometimes higher cloud tops, thin cloud-free regions paralleling the tracks, and often stronger radar returns. The ship-based aerosol measurements revealed that ship plumes often interacted with the overlying clouds in an intermittent rather than continuous manner. These observations imply that more must be learned about ship-track dynamics before simple relations between cloud condensation nuclei and cloud brightness can be developed.

1. Introduction This is due to the subtle boundary layer cloud pertur- bations that may trigger ship tracks and the dif®culty Ship-track clouds were ®rst described by Conover in measuring these perturbations. The marine environ- (1966) as anomalous cloud lines observed in satellite ment associated with ship tracks represents extremes images. These cloud lines can extend for hundreds of with respect to low concentrations of cloud condensa- kilometers and persist for several days. Multiple obser- tion nuclei (CCN) and the lack of surface temperature vations made from a small research vessel (R/V Glorita) during the Monterey Area Ship Tracks (MAST) exper- and roughness differences associated with convective iment in June 1994 are combined to describe the phys- turbulence effects in clouds. Numerical models that go ical and dynamic characteristics of ship-track clouds. A beyond most plume rise and dispersion models can be wide variety of aerosol and meteorological parameters useful in understanding the sensitivity of marine strat- were simultaneously measured from the R/V Glorita iform clouds to CCN and turbulence effects. Innis et al. with aircraft ¯ights. The focus of the surface, airborne, (1998, manuscript submitted to J. Atmos. Sci.) calculate and satellite studies during MAST was to improve the that temperature differences as small as 0.1ЊC can in- characterization of aerosol microphysical properties and crease cloud-level aerosol concentrations by over a fac- cloud dynamic processes in ship tracks (MAST 1994). tor of 2 in decoupled boundary layer conditions. Sys- Only a few studies of the in situ characteristics of tematic vertical velocities as low as a few centimeters ship tracks have been carried out. In studies to date, per second and/or associated air temperature increases emphasis has been on the aerosol microphysical char- of less than 1 K have produced modi®cations of marine acteristics of ship tracks (Ackerman et al. 1993; Al- boundary layer clouds in a statistical±dynamic numer- brecht et al. 1989; Radke et al. 1989; Ferek et al. 1998), ical model that mimic many of the morphological and while cloud dynamic aspects have been largely ignored. cloud liquid water characteristics associated with ship tracks (Porch et al. 1990). Important measurements were made during the MAST experiment from the R/V Glorita. Vertical pro- Corresponding author address: William M. Porch, Atmospheric ®les of background meteorological parameters (needed Physicist, D-407, Los Alamos National Laboratory, Los Alamos, NM 87545. as input to numerical models simulating ship tracks) E-mail: [email protected] were obtained from both rawinsonde and tethered bal-

᭧ 1999 American Meteorological Society

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC 70 JOURNAL OF APPLIED METEOROLOGY VOLUME 38 loons launched from the R/V Glorita (Syrett 1994). to about 500 m in the case of the USS Truxton and USS Also, surface properties such as sea surface tempera- Mount Vernon. Larger diesel within a few hundred tures and heat and moisture ¯uxes were obtained from kilometers of the USS Truxton and USS Mount Vernon measurements on the ship. Surface aerosol properties did make a visible ship track. More detailed information and lidar measurements of the interaction of ship plumes on these ships and their emissions is given by Hobbs and marine boundary layer clouds were made from the et al. (1998, manuscript submitted to J. Atmos. Sci.). ship (Hooper and James 1998, manuscript submitted to The instruments aboard the R/V Glorita consisted of J. Atmos. Sci.). Measurements of cloud bottom heights shipborne and balloon-borne sensors (Table 2). Ship- related to ship-track clouds were measured from the ship borne acoustic, optical, and microwave sensors probed with commercial . the boundary layer far from the in¯uence of the ship Research vessel±based measurements of atmospheric and high surface winds. In situ shipborne measurement parameters associated with ship tracks have several ad- systems included a tower for measuring heat and water vantages over aircraft measurements, such as the re- vapor ¯uxes, coupled with a ¯oating sea surface tem- search vessel being stationed for long periods at sea. perature measuring device, plus visible and infrared ra- Consequently, data were obtained on the evolution of diometry operated by the NOAA Environmental Re- conditions leading to ship-track formation during the search Laboratories (ERL) (Fairall et al. 1997). The tow- day and at night. A ship-based experiment, called SEA- er was located near the bow of the ship and was about HUNT, was performed in 1991 off the coast of Southern 6 m above the deck (10 m above sea level) to avoid as California and northern Mexico to study ship tracks and much as possible the effect of the ship's air wake. The other external forcing on marine boundary layer clouds instrumentation included global positioning system (Hindman et al. 1994). This experiment documented the (GPS) position detection and mast motion sensors to ®rst surface observation of a ship-track cloud that was account for translation and rotation effects on the hor- known to be a ship-track cloud simultaneously observed izontal wind and vertical velocities needed for eddy cor- by satellite. relation and estimates of bulk heat and water vapor In this paper, we report on the in situ and remote ¯uxes. measurement system results aboard the R/V Glorita dur- In situ measurements of aerosol characteristics were ing the MAST experiment. Data from these sensors are made using a sampling on the ship's mast connected to combined to characterize the physical and dynamic instrumentation in a cabin 20 m below. In situ mea- structure of ambient and ship-affected clouds. surements of aerosol characteristics were made using a 5-cm-diameter sampling tube 6 m up the ship's mast connected to instrumentation in a cabin 14 m below. 2. Ship-based measurements during MAST The analysis of the aerosol sampled from the ship's mast The most important component of the ship-based included condensation nuclei (CN), CCN, spectra (Hud- measurements in MAST was the research ship and its son and Li 1995), and aerosol size distributions from deployment with respect to dedicated navy ships and diffusion battery measurements. A limited number (12) ships of opportunity that affected marine stratiform of aerosol ®lter samples were taken from a sampler at- clouds during MAST. A complete description of the tached to the ship's high mast. A also was aircraft and ship operations, a description of the physical located on the mast. and power plants including the trajectory of each of the In situ balloon-borne sensors measured meteorolog- dedicated ships, and what is known about the ships of ical parameters. These included conventional tempera- opportunity can be found in Gasporivic (1995). In this ture, humidity, and wind pro®ling sensors for rawin- paper we will focus on the dedicated navy ships USS sonde and tethered balloons. Rawinsondes were Safeguard and USS Mt. Vernon; and on ships of op- launched every 3 h during potential ship track periods. portunity on 12, 27, and 28 June when ship tracks and Active remote sensing instruments included a scan- clouds that had been affected by ships that passed over ning lidar, which was mainly dedicated to studies of the research vessel. Figure 1 shows the ship tracks as ship plume and background aerosol inhomogeneities detected in the National Oceanic and Atmospheric Ad- and their transport and interaction with boundary layer ministration/Advanced Very High Resolution Radiom- clouds (Hooper and James 1998, manuscript submitted eter (NOAA/AVHRR) channel 3 satellite images for to J. Atmos. Sci.). Two commercial lidar±ceilometers these three days. An overview of information on the were mounted to the deck for continuous determination ships that produced many of the features in Fig. 1 are of cloud bottom heights and backscatter. The Pennsyl- given in Table 1. These ships varied in fuel type and vania State University (PSU) system was pointed ver- amount consumed from diesel, steam turbine, and nu- tically and the Los Alamos National Laboratory (LANL) clear. The nuclear ship, USS Truxton, and the steam system was oriented 40Њ from vertical pointing aft of turbine, USS Mount Vernon, did not produce a ship the ship. This was done to increase the path in the track. The relatively small USS Safeguard diesel ship boundary layer for increased aerosol sensitivity and to did make a ship track. The boundary layer depth was improve resolution of cloud edges during cloud open- about 300 m in the case of the USS Safeguard compared ings. A Doppler acoustic sounder (Porch et al. 1988)

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC JANUARY 1999 PORCH ET AL. 71

FIG. 1. NOAA AVHRR satellite images on 12, 27, and 28 June 1994 showing locations of R/V Glorita, dedicated navy ships, and ship tracks. These experiments were conducted 100±200 km west of Monterey Bay, CA. The dark features on 12 June are associated with cirrus clouds that were tending to mask the ship tracks on this day. The ship track from the USS Safeguard on 12 June was directly related to the navy ship and showed strong effects on the overlaying cloud morphology. On 27 June three prominent older ship tracks were observed that displayed smaller morphological effects. On 28 June the ship tracks were not as bright and were associated with larger commercial vessels rather than dedicated navy ships that used steam turbine and nuclear power sources.

TABLE 1. Ship characteristics.

Plume heat Speed Length (m)Ϫ displacement Ship Propulsion (MW)* (kt) (kg ϫ 106) USS Safeguard Diesel 1.6 14 77.7 Ϫ 2.6 4200 hp USS Mount Vernon Steam turbine 8.9 22 168.6 Ϫ 12.4 24 000 hp (Steam) USS Truxton Nuclear (Water cooled) 30 171.9 Ϫ 8.3 70 000 hp Tai Hi Diesel 8.2 19 231.6 Ϫ 32.6 22 088 hp Evergreen Diesel 8.1 20.5 230.8 Ϫ 33.5 Evergenius 21 600 hp

* Assuming 50% ef®ciency.

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC 72 JOURNAL OF APPLIED METEOROLOGY VOLUME 38

TABLE 2. Speci®cations of instruments used on the R/V Glorita. Accuracy Precision Instrument system Parameters Source (Ϯ) (Ϯ)* Meteorological (LANL) Cloud base height Handar 15 m N/S Ceilometer (PSU) Cloud base height VaÈisaÈlaÈ 15 m N/S Cloud video (time lapsed) Cloud ®eld images LANL N/S N/S Doppler radar Radar return LANL N/S N/S (LANL) 35 GHz Vertical velocity 5 cm sϪ1 1cmsϪ1 Doppler radar (PSU) 94 GHz Radar return PSU 0.2 dB/MSD 0.2 dB/MSD Doppler Horizontal, vertical ve- Remtech 0.2 m sϪ1 0.1msϪ1 2 Ϫ1 Ϫ1 locity and CT pro®les 5cms 1cms N/S N/S Flux tower (NOAA) Wind, temperature, and NOAA 0.2 m sϪ1 N/S heat ¯ux** 0.1ЊC N/S 10WmϪ2 N/S Lidar (Nd:YAG) Elastic aerosol back- NRL N/S 7.5 m scattering pro®les at 1.06 ␮m Microwave radiometer Vertically integrated PSU 0.13 cm 0.06 cm water vapor and liq- 40.7 ␮m 14.7 ␮m uid water path (LANL) Global solar radiation LiCor 10 W mϪ2 1WmϪ2 Pyranometer (NOAA) Global solar radiation Eppley 5 W mϪ2 0.5WmϪ2 Rawinsonde Temperature, wind, and VaÈisaÈlaÈ 0.2ЊC 0.1ЊC humidity pro®les 0.5msϪ1 0.2msϪ1 3% RH 1% RH Sea surface temperature sen- Temperature NOAA 0.2ЊC 0.1ЊC sor Temperature, wind, and AIR 0.5ЊC 0.05ЊC humidity pro®les 0.25 m sϪ1 0.1msϪ1 5%RH 0.1% RH Aerosol CN counter Condensation nucleii TSI 10 cmϪ3 1cmϪ3 CCN spectrometer Cloud condensation nu- Desert Research Insti- 10 cmϪ3 1cmϪ3 clei vs supersatura- tute 0.02% SS 0.01% SS tion Nephelometer Light scattering coef®- Radiance Engineering 2 ϫ 10Ϫ5 mϪ1 1 ϫ 10Ϫ5 mϪ1 cient

* Accuracy is based on estimates of variability of calibration vs a standard, while precision is based on the resolution of the variation. ** For more details, see Fairall et al. (1996). N/S is not speci®ed. RH is relative humidity. SS is supersaturation. was used for pro®ling wind, vertical velocity, and and time-lapse video system provided a continuous rec- boundary layer heights. This system could only be op- ord of cloud cover during the day. erated during rare periods of calm sea conditions (three days) because internal software disallowed data that dis- played too much variability. During calm periods, the 3. Results sounder was able to track wind speeds and vertical ve- a. Meteorological pro®les locities to about 700 m above the instrument. Two Doppler radars were aboard the ship. The PSU system Some of the most important measurements needed was a high-power pulsed system with a frequency of 94 for the dynamic modeling of ship-track clouds are tem- GHz. The LANL system was a small, battery-operated, perature and humidity pro®les. These were obtained on continuous wave (CW) system with a frequency of 35 a regular basis from rawinsonde launches from the ship GHz that was gimbaled to compensate partially for ship (every 3 h during cloudy periods). Table 3 summarizes motions. This system provided only integrated velocity the general sky conditions, whether a ship track was and return signal strength analysis. observed within about 500 km of the R/V Glorita, and Passive remote sensing instruments included a mi- an estimate of the marine boundary layer depth. The crowave radiometer (used to determine vertically inte- marine boundary layer depth often lifted considerably grated water vapor and cloud liquid water content), two through the day as weather systems passed, and at times , and a . A whole-sky camera as coastal processes affected the boundary layer.

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC JANUARY 1999 PORCH ET AL. 73

TABLE 3. Sky, ship track, and marine boundary layer depth esti- b. Surface conditions mates based on rawinsonde launches from the R/V Glorita (usually at 3±h intervals when clouds were present). Table 4 shows the surface conditions combined with Date Boundary layer pro®le parameters for the periods during three days (June 1994) Sky conditions Ship tracks depth (m) (with speci®ed time periods) when ship tracks (observ- 6 Clear No 500±1200 able by satellite) and ship-affected clouds (observable 8 Low clouds Yes 150±250 effects only from the surface) were observed passing 10 Fog No data 50±200 over the R/V Glorita. The surface conditions include 11 Low clouds Yes 200±450 data taken from the bow-mounted meteorological tower, 12 Low clouds Yes 350±400 sea surface temperature probe, and CN measurements. 13 Low clouds Yes 300±700 14 Clear No 450±650 All the data from the meteorological tower were cor- 15 Clear No 550±650 rected for ship-relative motion. The latent and sensible 16 Low clouds No 600±1300 heat ¯ux estimates are based on mean values of co- 17 Low clouds No 700±1300 variance, inertial dissipation, and bulk ¯ux estimates 20 Low clouds No 600±700 21 Low clouds No 350±600 (Fairall et al. 1996). The accuracy of these ¯ux mea- 2 22 Low clouds No 400±650 surements of about Ϯ10 W mϪ must be considered in 23 Clear No 200±500 the context of these mean ¯uxes and the importance of 26 Clear No 200±450 surface ¯uxes in numerical stratiform cloud models. The 27 Low clouds Yes 350±600 solar radiation data were derived from a diffusing-disk 28 Low clouds Yes 500±600 29 Low clouds Yes 400±500 pyranometer. The exposure for this instrument was not ideal, so the absolute irradiance is suspect (the primary radiation sensing system was destroyed early in the ex- periment during high seas). Table 3 shows that boundary layer depths less than Lidar measurements, scanning in both the horizontal and vertical, were used to describe the three-dimensional about 500 m were most conducive to ship-track obser- ship plume evolution from the ship into the marine vation. However, there were cases when the boundary boundary layer clouds. Figure 3 shows an example of layer depths were close to 500 m and no ship tracks this analysis for the USS Truxton plume on 28 June were observed near the R/V Glorita (e.g., 21±22 June). 1994. The USS Truxton was a nuclear-powered ship so Boundary layer depth is not the sole condition for ship- the measured plume was a combination of small ship- track observation. The population of ships in the vicinity board ef¯uents and sea spray generated by the ship's large enough or dirty enough to make a ship track is motion through the . One of the important results also important. A satellite survey conducted by Coakley of this analysis is that the plume rise was not continuous et al. (1998, manuscript submitted to J. Atmos. Sci.) but was strongly affected by convective elements that documents the importance of cloud height to the ob- are associated with the marine stratus clouds above. This servation of ship tracks during MAST. observation shows that overlaying clouds can affect On four days during the MAST experimental period, plume dispersal and entrainment of the plume into the the winds were low enough to permit tethered balloon clouds. Most of the plume rise seems to occur in regions launches from the ship for comparison with the rawin- where the surface level horizontal scan shows relative sonde pro®les. We were able to pro®le continuously minima in aerosol backscatter, indicating that the dy- from about 1100 to 1600 PDT 11 June. About ®ve ship namics of the clouds affect the entrainment of the aero- tracks were observed in the AVHRR satellite image on sol below. Details of plume rise times and diffusion are this day in clouds similar to those observed near the described in Hooper and James (1998, manuscript sub- ship. Unfortunately, none passed over the R/V Glorita. mitted to J. Atmos. Sci.). On days when there were no Figure 2 shows the short-term effect of a fast (15 min) overlaying clouds, the plume rise was more continuous. change in boundary layer height associated with an ap- The cloud is a balance of radiative cloud- proaching front and a change from a southerly to north- top cooling, latent heat release in cloud formation, and erly wind. This event followed a coastal surge the pre- surface heating. vious day (Nuss 1995). A coastal surge is a phenomenon observed by satellites re¯ecting the rapid movement of c. Cloud morphology and dynamics stratiform clouds up the coast of California. The asso- ciated changes in these clouds affected the conditions The properties of clouds affected by ships were re- under which ship tracks were observed during this pe- motely sensed with the ceilometers and Doppler radars. riod. The rawinsonde pro®les are included in Fig. 2 for Figure 4 shows the raw return signal for the tilted ceil- comparison. These show a similar change in boundary ometer during the periods that ship tracks passed over- layer depth after 1300 PDT but with considerably less head of the R/V Glorita on 12, 27, and 28 June. The vertical resolution than the tethered balloon measure- ceilometer laser could not penetrate most of the clouds ments. observed so only the cloud bottoms (shown in white)

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC 74 JOURNAL OF APPLIED METEOROLOGY VOLUME 38

FIG. 2. Comparison of (a) tethersonde and (b) rawinsonde pro®les during a coastal wind surge on 11 June 1994. This ®gure shows the rapid change in boundary layer structure and the extremely low cloud layer associated with the coastal surge and interaction with a moving front. are quantitative. The top of the white regions are more radiation at the sides of the ship track (especially on 12 a measure of the optical depth in the cloud rather than June). the cloud-top heights. Relatively strong effects on cloud The changes in cloud base features may be due to morphology were detected on 12 June, with weaker either cloud dynamic or cloud droplet microphysics. Ad- changes associated with ship-affected clouds the other iabatic cooling, either from the initial convective plume two days. Most of the changes are associated with the from the ship or from evolved dynamic convection in base of the clouds. The bottoms were lower by about the ship track, could explain the lowering of cloud-base 50 m than the background clouds. This condition has height associated with the ship track. Dynamic effects characteristics similar to the interaction of cooling tower are consistent with the thinner cloud regions at the sides plumes with overlaying clouds over land. On the other of ship tracks if general lofting in the ship track is com- hand, two cases on 27 June seem to show a slightly pensated for by subsidence at the edges. It is also pos- raised cloud bottom (about 20 m) associated with older sible that the lower parts of the cloud are related to tracks (the Tai He at about 1100 and an older track at drizzle. However, this would require drizzle production 1540 PDT). Supporting pyranometer data are also rather than suppression associated with the ship tracks shown in Fig. 4. The pyranometer data generally show that showed a cloud-base lowering. a relative decrease in solar radiation associated with the The small rises in cloud-base heights in the ship tracks ship track and often regions of relatively stronger solar on 27 June is a more subtle effect than the lowered

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC JANUARY 1999 PORCH ET AL. 75

TABLE 4. General conditions during periods of ship-track passes over research vessel or ship-modi®ed cloud observations. 12 Jun 1994 27 Jun 1994 28 Jun 1994 Parameters (units) (1300±1800 PDT) (1030±1430 PDT) (0630±1300 PDT) Wind speed at 10 m (m sϪ1) 9.7 9.6 8.9 Wind direction (degrees from true N) 315 343 347 Sea surface temperature (ЊC) 14.6 15.6 14.8 Air temperature at about 10 m (ЊC) 13.7 15.3 14.5 Speci®c humidity at about 10 m (g kgϪ1) 9.1 9.1 9.0 Sensible heat ¯ux (W mϪ2) 16.5 11.7 12.3 Latent heat ¯ux (W mϪ2) 28.2 51.6 11.6 Boundary layer depth (m) 400 600 550 Boundary layer potential lapse rate (ЊCkmϪ1) 2 1 1 Cloud base height (m) 180a 430b 330a Solar radiation (W mϪ2) 700 [1300±1500] 300±500b 100±400b [Times PDT] 300 [1500±1800] [1000±1400] [0830±1100] Condensation nuclei (cmϪ3) 1150 1000 620 Background 1100c 500 420 Ship plume 1500±10 000 900±1000 620±4800

a Lowering values throughout period. b Rising values throughout period. c Probable contamination of sampling line. cloud base observed near the origin of the ship track. cloud on 12 June (1444±1504 PDT associated with the Analysis from a three-dimensional statistical dynamic USS Safeguard). However, the ship track feature start- stratiform cloud model (Porch and Kao 1996) indicates ing at 1206 PDT 28 June showed raised clouds only at that once ship tracks are formed the combination of the edges, and the track associated with the Evergreen cloud radiative effects and the latent heat released as Evergenius on 28 June showed only a slight rise in the the cloud develops causes the cloud bottom height to center. Each of these clouds seemed as mature as the gradually rise with time. The ship tracks observed on two tracks with raised cloud bases on 28 June. The 27 June were more mature than the second ship-affected pyranometer data show a 30% or greater insolation on

FIG. 3. Lidar image showing ship-ef¯uent intermittent transport to cloud layer above the ship (28 June 1994). The discontinuous plume elements and the correspondence of clearer regions in the horizontal scan below the raised plume features demonstrates the involvement of the plume with active connective elements in the cloud layer above the plume.

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC 76 JOURNAL OF APPLIED METEOROLOGY VOLUME 38 her than the background clouds ship tracks are fresher and the cloud for 12, 27, and 28 June 1994 and pyranometer comparison. White corresponds to the highest 1 Ϫ . 4. Ceilometer raw backscatter data arbitrarily scaled backscatter units in [10 000 sr km] IG F backscattering values (cloud), followed by red (increased aerosol and virga), followed by yellow (boundary layer), and blue (background). When the bottoms are lower (12(27 and June). 28 June), the ship track cloud is lower than background clouds. When the clouds are higher and older, the track features are hig

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC JANUARY 1999 PORCH ET AL. 77

FIG. 5. Processed cloud-ceiling comparison from two ceilometers on 12 June 1994. The PSU ceilometer was oriented vertically, while the LANL system was tilted at 45Њ. This accounts for the height differences and time-lagged correlation.

12 June than on 27±28 June. The insolation variability stronger radar returns. However, the crossing of the Mt. is also greater, and the clouds were less dense and more Vernon plume around 0845 PDT showed no peak in CN variable on 12 June. The greater insolation implies a and was associated with a decrease in radar return (pos- greater surface heat ¯ux, which is supported in Table 4. sibly drizzle suppression). The lack of a peak in CN To check whether the features shown in Fig. 4 may implies that at this measurement point the ship plume have been due to instrument artifacts, comparisons were did not mix down to the sea surface. The lack of a CN performed between the processed cloud heights from peak makes this event dif®cult to time precisely. This the PSU and LANL ceilometers; one such comparison crossing was also complicated by a ship-track crossing is shown in Fig. 5. As expected, the tilted ceilometer at about 0906 PDT. At times (but not always) the top (LANL) shows a higher cloud-base height than the ver- of the clouds were slightly raised during the periods tically pointed system (PSU). A time lag is introduced affected by ships. This is consistent with visual obser- because the two ceilometers pointed at regions of the vations of a ship-track cloud extending higher than sky separated by about 300 m on 12 June and by about background clouds during the SEAHUNT experiment 500 m on 27 June (about the boundary layer height). (Porch et al. 1996) and the Sanko Peace ship track ob- On 12 June the lag peaks at about 1 min with a cor- served during MAST (Durkee et al. 1998, manuscript relation greater than 0.8. On 27 and 28 June the cor- submitted to J. Atmos. Sci.). relation was about the same (0.8), with time lags as long Figure 7 shows an example of the LANL CW Doppler as 3 min. Both systems show drops in the cloud-base radar data for the second ship-affected cloud feature heights associated with ship tracks and thinner cloud (USS Safeguard) at 1444±1504 PDT 12 June. Data stor- regions on both sides of the tracks. The drop in cloud age capacity limited the observation periods for this bottom height associated with the ®rst cloud feature instrument to about 1 h. This ®gure compares the radar beginning at 1322 PDT is not as great in the processed return strength (arbitrary units), the integrated Doppler cloud height data as in the raw data shown in Fig. 4. vertical velocity (compensated for ship and horizontal This implies that this feature, as well as many others cloud motion), the liquid water path from the microwave on subsequent days, represented a relatively thin cloud. radiometer, the pyranometer signal (compensated for This is consistent with visual observations that the ship- sun angle), and the ceilometer estimated cloud height. track feature at 1322 PDT had thin scudlike clouds be- The CN measurements showed a very strong plume as- low the main cloud that were not associated with back- sociated with this feature with CN concentrations rising ground cloud features. to a maximum of about 10 000 cmϪ3. Concentrations The two Doppler radars also remotely sensed char- outside the plume were in the range of 1000 cmϪ3 (this acteristics of the ship tracks. Figure 6 shows an example was a very clean day, as observed by aircraft measure- of the data from the PSU system on 28 June. In general, ments and may indicate some sample contamination in when peaks in CN indicated ship plumes below ship- this case). The remote sensing comparison in Fig. 7 affected clouds these peaks seemed to be associated with shows that there is an apparent correspondence between

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC 78 JOURNAL OF APPLIED METEOROLOGY VOLUME 38

FIG. 6. The Pennsylvania State University Doppler radar re¯ectivities on 28 June 1994 to a height of 1 km above the sea surface. The highlighted times correspond to the ship-affected cloud observations shown in Table 5. This ®gure shows the variability of cloud radar re¯ectivities on 28 June and the lack of de®nite relationship between ship tracks (with associated increases in optical re¯ectivities) and decreased cloud radar re¯ectivities (actually, the re¯ectivities seemed to be larger except for the 0845±0850 PDT) expected if drizzle-sized droplets are diminished.

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC JANUARY 1999 PORCH ET AL. 79

comes from the fact that cloud effects were observed from the ship that did not correspond to a satellite- observed ship track. This is due either to a timing difference between the satellite images and ship over- pass or the feature was too small or too weak at cloud top to be resolved by satellite. Although, in general, these features were associated with increases in CN at ship level, this was not always the case. Many of the features showed decreases in pyranometer signal and increases in liquid water content and radar re- ¯ectivity, but again this was not always the case. As discussed previously, although the younger ship tracks and ship-affected cloud features usually dem- onstrated a slight lowering of cloud base, older ship tracks showed little effect or a slight rise. In spite of the small contribution of surface effects in generating convective turbulence in marine stratiform clouds, convective elements develop in these clouds and ship tracks embedded in them. This affects the measure- ments made in each pass underneath them at the sur- face by a ship as well as aircraft passes through them. These convective elements are associated with cloud microphysics, cloud morphology, and CCN concen- trations. Taken together, these data only suggest oc- casional dynamic and microphysical effects of ships on clouds. More measurements are needed to provide FIG. 7. LANL Doppler radar, MWR, and pyranometer comparison a robust statistical description of ship-track charac- for 12 June 1994. The vertical bars bound the ship plumes as detected teristics. in the CN data. These comparisons show the higher radar signal level and slightly increased vertical velocities from the Doppler radar and associated higher liquid water content from the microwave radiom- eter, decreased solar irradiance, and cloud bottom height associated 4. Conclusions with the smaller ship-track feature shown in Fig. 4 for 12 June. Ship-based measurements provided unique informa- tion related to ship-track clouds from surface measure- ments and meteorological pro®les from tethered and free the radar return strength and the cloud liquid water. balloons. The data are critical as inputs to, and con- Also, where the cloud liquid water is reduced at the straints on, numerical models designed to simulate the sides of the feature, the Doppler vertical velocities show effects of ship-plume aerosols, heat and moisture, and small peaks in subsidence. The transparency of the ship air wakes on marine stratiform clouds. Aerosol cloud, as shown by the pyranometer, peaks when the measurements of CN at ship-mast-level identi®ed ship cloud liquid water decreases at the sides of the feature; plumes. Lidar backscattering images showed ship the subsidence is maximized and the cloud bottom plumes couple with convective elements in overlaying height reaches its relative maximum. cloud layers. The ship track events on 12 June are the best example Information was obtained on cloud morphological obtained during MAST of comparisons between the changes associated with ship tracks. Comparison of ship-based measurements during a ship-affected cloud data from remote sensors aboard the ship shows that event. However, most of the features were observed at ships often affected the base and top heights of clouds. other times, although more subtly. Table 5 generalizes Comparison of data from remote sensors aboard the the comparisons for the ship-track and ship-affected ship shows that ships often affected the base and top cloud situations we have been able to identify for 28 heights of clouds by as much as 50±100 m. Relatively June. The CN concentrations at the surface were much strong effects on cloud morphology were detected on lower during the two ship-track passes on 28 June than 12 June, showing a drop in cloud base of 100 m in a on 12 June, re¯ecting the close location of the USS freshly produced ship track. Pyranometer measure- Safeguard on the 12th and the greater age of the ship ments showed increased solar irradiance values of tracks on the 28th. about 400 W mϪ2 on both sides of the ship track. Also, Table 5 shows the variability of characteristics ob- ship-affected clouds often have thin cloud or cloud- served from the ship of ship-affected clouds. The gen- free regions on their sides. These physical features eral separation of ship-track and ship-affected clouds indicate that cloud dynamics may often be an impor-

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC 80 JOURNAL OF APPLIED METEOROLOGY VOLUME 38 small peaks at 0708 and 0736 0826 and 0830 center 0828 peaks at 0900 and 0915 peaks at 0915 and 0924 at 1020 and 1050 fect crease through period Low light with Peaks at edges Slight decrease in Decrease with small Little effect peaks No observable ef- Slight or no de- crease in middle, dips at 0708 and 0736 LWC with dips at edges at 0826 and 0830 decreases at 0900 and 0910 LWC with dip at 1020 crease in LWC before/after 0845, decrease 0845± 0850 Slight increase in Very slight de- 1 Ϫ ) 1020 1 Ϫ nal increase subsidence (5 cm s peaks±25 cm s 1050, 1113, and 1120 No observable sig- Slight increase in No data No effect Very small peak No data LWC high with de- No data Noticeable peak in No dataNo data Small increase with No effectStrong subsidence Small decrease with No data Increase in LWC crease 0957± 1012 strong returns returns 0823± 0830 with de- crease at 0826± 0828 turns 0850±0910 decrease at 0902 turns 0915±0924 fect turns from 0835± 0845, weak from 0845±0855 (driz- zle suppression?) Weak signal in- Very weak increase 0806 Period of relatively Period of stronger Period of strong re- No observable ef- Period of strong re- (20 m) raw data 0958± 1020 drops 50 m drops at sides 0708 and 0736 dip 10 m at 0826 m at 0906 data, only one dip in ceiling 50 m at 1052 about 0845 Stronger signal in 20 m at 0802 No or very weak Strong dip 20±40 3 Ϫ back- ) Ceilometers PSU radar LANL radar MWR Pyranometer 3 3 3; Ϫ Ϫ Ϫ 4800 cm 0910 ground 420 cm 0820±0830 No data Dips on each side Short peak CN No data Variable ceiling Short peak 0906± CN 620 cm No peaks Strong dips in raw 5. Measurement comparisons during ship-affected cloud events on 28 Jun 1994. ABLE T 1052 image N12ch3 image ble N12ch3 1052 image Strong N12ch3 1052 image 2d Ship track N 12ch3 0928 image 14 min) 5 min Ϯ 08060820±0830 Not No visible visible on Short peak 0806 Very slight dip 0915±0924 Not visible0952±1007 ( No data Slight dip 20 m Period of strong re- Ϯ 0845±0850 Not visible No peak Slight dip 25 m Mt. Vernon Event Times (PDT) Satellite CN (cm Mt. Vernon Mt. Vernon Mt. Vernon fect fect * N12ch3 is NOAA AVHRR 9 Channel 3. fect (hard to time; no CN peak) fect Weak ship cloud ef- Weak ship cloud ef- Weak track 0708±0736 N12ch3* Weak ship cloud ef- Weak ship cloud ef- Ship Track Evergreen Evergenius Older track? 1050±1120 Only ®rst part visi- Ship track 0906±0914

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC JANUARY 1999 PORCH ET AL. 81 tant component of ship-track features. These features, Young, 1996: Bulk parameterization of air±sea ¯uxes for Trop- combined with the fact that ship tracks seem to occur ical Ocean±Global Atmosphere Coupled Ocean±Atmosphere Re- sponse Experiment. J. Geophys. Res., 101, 3747±3764. only within a relatively narrow range of boundary , A. B. White, J. B. Edson, and J. E. Hare, 1997: Integrated layer depths (about 0±600 m; Durkee et al. 1998, shipboard measurements of the marine boundary layer. J. Oce- manuscript submitted to J. Atmos. Sci.), present a anic Atmos. Technol., 14, 368±379. challenge to numerical modeling of ship tracks. The Ferek, R. J., D. A. Hegg, P. V. Hobbs, P. Durkee, and K. Nielsen, 1998: Measurement of ship-induced cloud tracks off the Wash- possibility that cloud droplet distributions may differ ington coast. J. Geophys. Res., in press. with height in ship-track clouds compared to back- Gasparovic, R. F., 1995: MAST Experiment Operations Summary. The ground clouds due to internal dynamics present a chal- Johns Hopkins University Applied Physics Laboratory, 295 pp. lenge to the interpretation of aircraft measurements Hindman, E. E., W. M. Porch, J. G. Hudson, and P. A. Durkee, 1994: of aerosol and droplet spectra. Ship-produced cloud lines of 13 July 1991. Atmos. Environ., 28, 3393±3403. Hudson, J. G., and H. Li, 1995: Microphysical contrasts in Atlantic Acknowledgments. This work was performed under stratus. J. Atmos. Sci., 52, 3031±3040. support from the Department of Defense, Of®ce of Na- MAST, 1994: Monterey Area Ship Track (MAST) Experiment: Sci- val Research Grant N00014-94-1-0681, and the De- ence plan. Naval Postgraduate School Rep., NPS-MR-94-004, 45 pp. partment of Energy, Of®ce of Health, and Environmen- Nuss, W., 1995: NPS Coastal Boundary Layer Experiment 1994. Na- tal Research Grant KP-01. The authors would like to val Postgraduate School, 68 pp. thank J. Archuleta, M. Buchwald, P. Hobbs, L. May, T. Porch, W. M., and C-Y Kao, 1996: Ocean measurements and models Najita, W. Shaffer, W. Spurgeon, B. Albrecht, D. Babb, of ship trail cloud characteristics. Second Int. Conf. on Global Energy and Water Cycle, Washington, DC, GEWEX, 353±355. C. Fairall, J. Hudson, G. Demme, D. Kaplan, M. Miller, , W. D. Neff, and C. W. King, 1988: Comparisons of meteoro- and the captain and crew of the R/V Glorita. logical structure parameters in complex terrain using optical and acoustic techniques. Appl. Opt., 27, 2222±2228. , C.-Y. Kao, and R. G. Kelley, 1990: Ship trails and ship-induced REFERENCES cloud dynamics. Atmos. Environ., 24A, 1051±1059. , , M. I. Buchwald, W. P. Unruh, P. A. Durkee, E. E. Hind- Ackerman, A. S., O. B. Toon, and P. V. Hobbs, 1993: Dissipation of man, and J. G. Hudson, 1996: The effects of external forcing marine stratiform clouds and collapse of the marine boundary on the marine boundary layer: Ship trails and a solar eclipse. layer due to depletion of cloud condensation nuclei by clouds. Global Atmos. Ocean Syst., 3, 323±340. Science, 262, 226±228. Radke, L. F., J. A. Coakley, and M. D. King, 1989: Direct and remote Albrecht, B. A., 1989: Aerosols, cloud microphysics, and fractional sensing observations of the effects of ships on clouds. Science, cloudiness. Science, 245, 1227±1230. 246, 1146±1148. Conover, J. H., 1966: Anomalous cloud lines. J. Atmos. Sci., 23, 778± Syrett, W. J., 1994: Low-Level Temperature and Moisture Structure 785. from the Monterey Area Ship Track Experiment. The Pennsyl- Fairall, C. W., E. F. Bradley, D. P. Rogers, J. B. Edson, and G. S. vania State University, Department of Meteorology, 35 pp.

Unauthenticated | Downloaded 10/02/21 06:02 PM UTC